Clustering of tactile maps for a single object interaction

This is an interactive version of Figure 2e in the paper. Hover your mouse cursor to explore the PCA embedding of a single object interaction from our dataset. Each data point corresponds to a single captured frame. The 5 groups correspond to 5 k-mean clusters used for sampling for our object classifier with 5 input frames. The right image and the tactile map show the data for the data point highlighted in the embedding map on the left side.